Improving compilation time - Revision history

Alissa Sabre: Added to Category:Compiling viewer.

2008-07-01T14:19:27Z

← Older revision		Revision as of 07:19, 1 July 2008
Line 12:		Line 12:

	=== Hardware ===		=== Hardware ===
	* ~~Athlon~~ 64 X2 5200+ (1MB cache)		* Athlon 64 X2 5200+ (1MB cache)
	* ~~4GB~~ ECC DDR2 666 RAM		* 4GB ECC DDR2 666 RAM
	* ~~Root~~ on RAID-1, SAMSUNG HD300LJ and Maxtor 6V300F0		* Root on RAID-1, SAMSUNG HD300LJ and Maxtor 6V300F0

	=== Settings ===		=== Settings ===
	* ~~Compiling~~ with -j2		* Compiling with -j2
	* ~~llmozlib~~ enabled		* llmozlib enabled
	* ~~building~~ 32 bit SL version		* building 32 bit SL version
	* ~~Source~~ from http://svn.daleglass.net/secondlife, rev 81.		* Source from http://svn.daleglass.net/secondlife, rev 81.

	=== Software ===		=== Software ===
	* ~~Gentoo~~, x86_64, gcc 4.1.1		* Gentoo, x86_64, gcc 4.1.1
	* ~~Ubuntu~~ Edgy, x86_64, gcc 4.1.2		* Ubuntu Edgy, x86_64, gcc 4.1.2

	=== Methodology ===		=== Methodology ===
Line 80:		Line 80:
	\|\| \|\| System \|\| 0m4.599s \|\| 0m4.360s \|\| 0m3.546s \|\| 0m3.258s		\|\| \|\| System \|\| 0m4.599s \|\| 0m4.360s \|\| 0m3.546s \|\| 0m3.258s
	\|}		\|}

			[[Category:Compiling viewer]]

Cron Stardust: /* Improving compilation time */

2007-04-26T03:12:14Z

Improving compilation time

← Older revision		Revision as of 20:12, 25 April 2007
Line 1:		Line 1:
	== Improving compilation time ==		== Improving compilation time ==
	Based on the benchmarks below, the following conclusions were made:		Based on the benchmarks below, the following conclusions were made:
	* The more RAM, the better. Enough RAM will ~~made~~ the hard disk's performance irrelevant.		* The more RAM, the better. Enough RAM will make the hard disk's performance irrelevant.
	* Using a RAM disk for /tmp or even the full system doesn't noticeably improve performance if enough RAM is present.		* Using a RAM disk for /tmp or even the full system doesn't noticeably improve performance if enough RAM is present.
	* Given enough RAM, the process will be entirely CPU bound.		* Given enough RAM, the process will be entirely CPU bound.
Line 8:		Line 8:
	* Gentoo compiled 15% faster.		* Gentoo compiled 15% faster.
	* Gentoo ran scons 50% slower.		* Gentoo ran scons 50% slower.


	== Benchmarks ==		== Benchmarks ==

Dale Glass: Benchmark results and conclusions

2007-04-01T15:25:14Z

Benchmark results and conclusions

New page

== Improving compilation time ==
Based on the benchmarks below, the following conclusions were made:
* The more RAM, the better. Enough RAM will made the hard disk's performance irrelevant.
* Using a RAM disk for /tmp or even the full system doesn't noticeably improve performance if enough RAM is present.
* Given enough RAM, the process will be entirely CPU bound.
* A better hard disk will make a significant difference in link time if enough RAM can't be obtained.
* Compiling using more processes (-j3 vs -j2) may be counterproductive if there's not enough RAM.
* Gentoo compiled 15% faster.
* Gentoo ran scons 50% slower.

== Benchmarks ==

=== Hardware ===
* Athlon 64 X2 5200+ (1MB cache)
* 4GB ECC DDR2 666 RAM
* Root on RAID-1, SAMSUNG HD300LJ and Maxtor 6V300F0

=== Settings ===
* Compiling with -j2
* llmozlib enabled
* building 32 bit SL version
* Source from http://svn.daleglass.net/secondlife, rev 81.

=== Software ===
* Gentoo, x86_64, gcc 4.1.1
* Ubuntu Edgy, x86_64, gcc 4.1.2

=== Methodology ===

Gentoo tests were done with /tmp on tmpfs.

Ubuntu tests were done with the whole install on /tmp on tmpfs, and the whole install on disk. This took about 1.5GB RAM:

* 334MB for temporary files generated during compilation
* 642MB for the source code
* 462MB for the Ubuntu install (probably can be trimmed a bit)

Ubuntu tests were done as follows: Distribution was installed with debootstrap. Sources were copied into it (same copy as used for Gentoo). The whole tree, OS and sources included was copied to /tmp (on tmps), and chrooted into.

The "(on disk)" tests were done afterwards, by copying the whole tree (1.5GB) back to disk (as some changes were required to get the build going).

The "(after reboot)" were done as follows:
# Do required preparations (eg, binary removal)
# Reboot
# Login
# Chroot
# Build

X wasn't running during the tests.

=== Results ===

{|{{Prettytable}}
|+'''Benchmark results'''
|-{{Hl2}}
!Test
!Timings
!Gentoo
!Ubuntu (after reboot)
!Ubuntu (on disk)
!Ubuntu (on RAM disk)
|-
|| Full build || Real || 16m44.594s || 19m23.391s || 19m12.447s || 19m5.572s
|-
|| || User || 29m25.549s || 35m15.595s || 35m23.223s || 35m24.639s
|-
|| || System || 2m57.599s || 2m9.673s || 2m16.113s || 2m13.387s
|-

|| Rebuild || Real || 0m41.752s || 0m27.656s || 0m15.457s || 0m15.444s
|-
|| || User || 0m39.938s || 0m14.858s || 0m14.963s || 0m14.771s
|-
|| || System || 0m1.722s || 0m0.952s || 0m0.644s || 0m0.784s
|-
|| Relink || Real || 1m24.408s || 1m17.616s || 0m41.092s || 0m41.268s
|-
|| || User || 1m12.687s || 0m37.895s || 0m37.455s || 0m38.016s
|-
|| || System || 0m4.599s || 0m4.360s || 0m3.546s || 0m3.258s
|}